502

E. W. BETH

theoretical investigation; for the philosophy of (exact) science, in particular, there is no reason to investigate certain subtleties· which present themselves·- in natural language and which are usually avoided in the construction of fonnalised languages. I feel these points of agreement should be explicitly stated, as, naturally, in the above considerations they did not receive the stress which they deserve. I now come to the issues on which Carnap·s· views and mineseem -to diverge. (iv) In discussions on the foundations of logic, natural language plays a special role in those cases, where strict usage is desired; though there is no reason to exclude strict usage of formalised languages, strict usage is actually restricted to natural language; this is also the case in Carnap's writings; (v) His neglect of the distinction between strict usage and amplified usage of a language has induced Carnap to defend assertions-and, in particular, the Principle of Tolerance-which -cannot be ac~ cepted without restrictions; moreover, Carnap has not been able to avoid every appeal to logical or mathematical intuitions, or, what amounts to the same, to ontological commitments. This criticism leaves the main body .of Carnap's doctrines fully intact. It only calls for modifications and restrictions at its ultimate boundaries, which a:t the same time are the boundaries of our knowledge. E. W. INSTITUUT VOOR GRONDSLAGENONDERZOEK EN PHILOSOPHIE DER EXACTE WETENSCHAPPEN, UNlVERSITEIT VAN AMSTERDAM

BETH

16 P. F. Strawson

CARNAP'S VIEWS ON CONSTRUCTED SYSTEMS

VERSUS NATURAL LANGUAGES IN ANALYTIC PHILOSOPHY The Two Methods

I

,;

understand that the question on which I am to discuss Carnap's f views is that of the comparative merits of two methods of philosoph-

ical clarification. To follow one method is to construct a formal system,

which uses, generally, the ordinary apparatus of modern logic and in which the concepts forming the subject-matter of the system are introduced by means of axioms and definitions. The construction of the system will generally be accompanied by extra-systematic remarks in some way relating the concepts of the system to concepts which we already use in an unsystematic way. This is the method of 'rational reconstruction';

and indeed the system of elementary logic itself can be regarded as just such a reconstruction of the set of concepts expressed by the logical constants of daily life. Following the other method seems very different. For it consists in the attempt to describe the complex patterns of logical behaviour which the concepts of daily life exhibit. It is not a matter of prescribing the model conduct of model words, but of describing the actual conduct of actual words; not a matter of making rules, but of noting

customs. Obvionsly the first method has certain advantages. The nature and powers of the apparatus to be used are clear. Its users know in advance what sort of thing they are going to make with it. The practitioner of the second method is not so well placed. Unless he is to be content with the production and juxtaposition of particular examples, he needs some metavocabulary in which to describe the features he finds. Ex hypothesi, the well-regulated metavocabulary of the first method is inadequate for his purposes. So he has to make his own tools; and, too often, hastily improvised, overweighted with analogy and association,

they prove clumsy, lose their edge after one operation and serve only to mutilate where they should dissect.

Clarification and Science The issue, or apparent issue, between the two methods is only too 503

504

P. F. STRAWSON

easily trivialised or made uninteresting. I spoke of them both as methods of clarification, and one could understand this word in such a way that there was no interesting question as to which of the two methods was better for this purpose. Such a result would ensue, for ~xamp~e,. from taking 'clarification' in. the sense which CaTnap seems to gIve to It In ~he first chapter of 'Logical Foundations of Probability'.' A pre-scientIfic concept C is clarified in this sense if it is for ce~tazr: pUTfoses ~epla~ed (or supplanted or succeeded) by a concept C' whICh IS unhke C m bemg both exact and fruitful. The criterion of exactness IS that the rules of use of the concept should be such as to give it .a clear place .';n a wellconnected system of scientific concepts'. The cnten~n of frultfulnes~ IS that the concept should be useful in the formulatIOn of many logIcal theorems or empirical scientific laws. An indication of the sense In which the new concept is said to correspond to and to replace the old may be given by examples. One example which Carnap gives ~s ~he replacement of the sensory quality concept of warmth by the quantItatIve concept of temperature. An example analogous. to anoth~r whI~~ he gives would be the use by the entomologist of the word msect m a way more restrictive and more exactly define~ than th~ way (or ways) in which it is used by children and nursemalQS.. There 15 a further suggestion (though not an explicit assertion) in thIS chapter, w the effect that introducing a concept into a well-connected sy.sten:- of sCIentific cone tS and constructing a formal axiom system whIch Incorporates both cit pand them, are really just different names f or th e same tho . Ing. 2 A n d if 'clarification' is so understood as to include 'rende.ring.exact', and 'rendering exact' is understood to ,include incorporatIOn In a fo~al system, then clearly the thesis that clarification can be best achIeved by system-construction appears as an under.statement.. . Even if we abjure this last step and thmk of clarlfica~IOn.more vaguely as the introduction, for scientific purposes, of SCIentifically exact and fruitful concepts in the place of (some of) t~ose we ~se for all the other ordinary and extraordinary purposes of hfe, the Issue between the two methods remains less than exciting. I am not competent to discuss the extent to which theoretical scientists, in framing new c?ncepts or refurbishing old ones, either examine minutely the behaVIOur of words in ordinary language or construct axiom systems. It seems to me extremely improbable that they do much of the first; and I suspect (but may be quite wrong) that logicians exaggerate the extent to. whIch they do, or ought to do, the second. But my mc?mpetence m thIS m~t­ ter troubles me not at all. For however much or lIttle the constructIOnIst technique is the right means of getting an idea into shape for use in the lef. op. cit., 3-15. 2See 15.

CARNAP ON CONSTRUCTED SYSTEMS

505

formal or empirical sciences, it seems prima facie evident that to offer formal explanations of key terms of scientific theories to one who seeks philosophical illumination of essential concepts of non-scientific discourse, is to do something utterly irrelevant--:-is a sheer misunderstanding, like offering a text-book on physiology to someone who says (with a sigh) that he wished he understood the workings of the human heart. The scientific uses of language, whether formal or empirical, are extremely highly specialized uses. Language has many other employments. We use it in pleading in the law courts; in appraising people's characters and aetions; in critidsing works of art; in recounting our states of mind; in getting people to fetch things; in narrating histories; in describing what things look and sound and feel like; in entering into engagements with one another; in identifying people-and so on. It is quite certain that such ways of using language as these may give rise to philosophical problems; that the concepts employed in these activities may generate perplexity, may call for philosophical clarification. How do we conceive of responsibility? What is the difference between describing states of mind and describing physical objects? What does ,it mean to say that the person now before us is the very man who did such-and-such a thing? This is a minute and random selection of typical questions concerning concepts employed in non-scientific discourse. Moreover the language used outside the research institutes has ilts general and structural features, running through quite disparate realms of subject-matter and purpose. These too have seemed to demand philosophical investigation. Thus we wish to know what it is to say that ,one thing is conditional upon another, is a case or instance of another, is real, is good, is the same. And it seems in general evident that the concepts used in non-scientific kinds of discourse could not literally be replaced by scientific concepts serving just the same purposes; that the language of science could not in this way supplant the language of the drawing-room, the kitchen, the law courts and the novel. It might at this point be objected that while it is trivially true that doing science is not doing not science, it does not follow that the employment of scientific con~ cepts for the purposes for which non-scientific concepts are at present employed is impossible; i.e. that from the necessary truth that scientific uses of language are different from non~scientific uses of language, it does not follow that use of scientific language could not replace the use of non-scientific language for non-scientific purposes. And of course it does not follow; and, in certain. cases, for certain descriptive purposes, the replacement might be effected. But it seems to require no argument to show that, in most cases, either the operation would not be practically feasible or the result of attempting it would be something so radically different from the original that it could no longer be said to be fulfilling the same purpose, doing the same thing. More of the

;

506

P. F. STRAWSON

CARNAP ON CONSTRUCTED SYSTEMS

types of linguistic activity in· which we constantly engage would succumb to such an attempt than would survive it; and there are many such types on which we should uot know how to start. The kinds of coucept we employ are uot indepeudeut of the kiuds of purpose for which we employ them; eveu though some concepts can fulfil more than one kind of purpose. lf these things are true, it follows that typical philosophical problems about the concepts used in non-scientific discourse cannot be solved by laying down the rules of use of exact and fruitful concepts in science. To do this last is not to solve the typical philosophical problem, but to change the subject. In the case of many a philosophically troubling concept, indeed, it is hard to know in what direction to look for a scientifically satisfactory concept which stands to it in the required relation of correspondence or similarity. But the general conclusion holds even for those cases where there is a clear correlation. I may mention again Carnap's own example of the clarification of the prescientific concept of warmth by the introduction of the exact and scientifically fruitful concept of temperature. Sensory concepts in general have been a rich source of philosophical perplexity. How are the look, the sound, the feel of a material object related to each other and to the object itself? Does it follow from the fact that the same object can feel warm to one man and cold to another that the object really is neither cold or warm nor cool nor has any such property? These questions can be answered, or the facts and difficulties that lead to our asking them can be made plain; but not by means of formal exercises in the scientific use of the related concepts of temperature, wavelength, frequency. Indeed, the introduction of the scientific concepts may itself produce a further crop of puzzles, arising from an unclarity over the relations between two ways of using language to talk about the physical world, the relations between the quantitative and the sensory VO~ cabularies. This unclarity is another which wiU scarcely be removed by exhibiting the formal workings of the quantitative concepts. So, then, since the clarification of philosophically puzzling concepts is not, achieved by the introduction of related scientific concepts, it is not important for our purpose to discuss whether this introduction is best performed by the method of formalisation. Nor is it very important to discuss this, even if 'clarification of philosophically puzzling concepts' is taken to be synonymous with 'introduction of related scientific concepts'. For the answer is trivially 'Yes' for formalised sciences and trivially 'No' for others. lf it is objected that the real question is whether all sciences would. not be the better for axiomatisation, then the real question is one which I must leave, thankfully, to those equipped to answer .it.

Clarification and 'Pseudo-questions'

507

It is possible, however, to understand the idea of clarification, and 0: the contribution which may make to in a system~construction

it,

dIfferent and more philosophical way; in such a way, in fact, that the issue stated at the outset remains open, requires to be argued further. ~ut before I turn to this other approach, I want to consider a posSIble source of the uncompromising pOSition I have just discussed. I think it arises partly from the view that philosophical questions and perplexities cannot realI.y be taken seriously; that the only serious questions are elther questIOns to be answered within the conceptual framework of a scie.ntific. theory ?r of some nonwscientific mode of empirical discourse pragmatIc questIons about the desirability of adopting such a frame.! :"ork.. ~~is view is ~trongly suggested by certain passages in Carnap's EmpIrICISm, SemantIcs and Ontology'.3 It is worth while to consider in det~ some of the things which Carnap says in this article, since to do so. WIll both illuminate our general question, and show how thin (despIte appearances) may be the barrier which divides the philosopher who constructs systems from the philosopher who describes the workin?~ of or~inary language. Camap declares that the 'framework of propoSllIDns' (I.e. a use of language in which propositions appear as entities having such properties as truth, necessity, etc.) may be introduced by means of a set of rules, of which he indicates a few. Then he continues as follows: 4

or,;

(~) It is important to notice that the system of rules for the linguistic exw presslOns of the propositional framework (of which only a few rules have here been briefly indi~ated) is sufficient. for the introduction of the framework. Any further explanations as to the nature of the propositions (i.e. the elements of the framework indicated, the values of the variables up", "q", etc.) are theow retically unnecessary because, if correct; they follow from the rules. (ii) For example, are propositions mental events (as in Russell's theory)? A look at the rules shows us that they are not, because otherwise existential statements would be of the form: "If the mental state of the person in question fulfils such-andwsuch conditions, then there is a p such that . . . >t. The fact that no references to mental conditions occur in existential statements [of the framework] shows that propositions are not mental entities. Further, a state. ment of the existence of linguistic entities (e.g. expressions, classes of expressions, etc.) must contain a reference to a language. The fact that no such reference occurs in the existential statements here, shows that propositions are ~~evue . lnternationale de PhilosophieJ XI (1950), 20-40. But the article is itself a br~hant mformal contribution to the philosophical clarification of the concept of

exIstence. How would Carnap charactense what he himself is mainly concerned to do in the article?

40p. cit.J 26-27. I quote a continuous paragraph, which I have broken up into numbered sections -for ease of reference. .

508

P. F. STRAWSON

CARNAP ON CONSTRUCTED SYSTEMS

not linguistic entities. The fact that in these statements no reference to a subject (an observer or knower) occurs (nothing like "There is a p which is necessary for. Mr. X"), shows that the propositions (and their properties, like necessity,

concerned, who have no practical. difficulties at all in operating with the concepts in question. To the extent to which Carnap regards the role of extra-systematic conceptual explanation as simply that of resolving such practical difficulties, he ignores the role of conceptual explanation in resolving philosophical difficulties; and this perhaps springs again from the view that the latter are not real difficulties. And of course they are not (in general) real difficulties, if by 'real difficulty' is meant a difficulty in actually operating with the concepts in question in the course of framing and answering 'real questions', i.e. questions which arise within the framework to which the concepts belong. (3) Carnap says that the extra-systematic explanations are analogous, ~ to the pictorial models by means of which scientists may introduce the-; oretical concepts to a beginner (and, he might have added, which they' may themselves make use of in extending and applying their theories). But it is easy to see that they are not analogous, just in so far as the conceptual explanations 'follow from', or are implicit in, the rules of use of the expressions concerned. For it does not appear to be the case, indeed it is not clear what would be meant by saying, that the scientists' pictorial models 'follow from' any 'rules of use' of the relevant scientific expressions. Another respect in which the two things are not analogous is that the scientists' models do seem to be of practical use in helping the beginner to learn to use the theoretical concepts in question and, perhaps, in helping the scientist to frame and extend theories; whereas the explanations which are of help to the philosopher do not in general have, or need, this power. It seems not unreasonable, then, to find in this passage, as in others~ evidence of a lack of sympathy with, and even of understanding of, that need for the elucidation of concepts which can coexist with perfect mastery of their practical employment. Now this is precisely the need for their philosophical elucidation. But if the idea of this kind of clarification is rejected, or not even entertained, then it does become intelligible that the title of 'clarification' should be reserved for some other activity. And this is why I said that a certain extreme view of the nature of clarification is perhaps traceable in part to the belief that philosophical questions and difficulties' are non-serious and unreal. This was the ex~ treme view that to clarify a concept used for non-scientific purposes consisted in looking away from it at a different, though in some way related, concept which was unlike the first in being scientifically exact and fruitful. It is true that we may be diverted from the wish to understand what we are doing, by encouragement to do something else; and that if the wish seems futile, the diversion may seem desirable; and then the complaint that the wish is not thereby satisfied will, no doubt, seem futile too.

etc.) are not subjective. (iii) Although characterisations of these or similar kinds are, strictly speak. iug, unnecessary, they may nevertheless be practically useful. If they are given,

they should be understood, not as ingredient parts of the system, but merely as marginal notes with the purpose of supplying to the reader helpful hints or convenient pictorial associations which may make his learning of the use of the expressions easier than the bare system of rules would do. (iv) Such a characterization is analogous to an extra-systematic explanation which a physicist sometimes gives to a beginner. He might, for example, tell him to imagine the atoms of a gas as small balls rushing around with great speed, or the electro-magnetic field and its oscillations as quasi-elastic tensio~s and vibrations· in an ether". In fact, however, all that can accurately be saId about atoms or the field is implicitly contained in the physical laws of the theories in question.

It will be noticed that the existence of typical philosophers' questions about propositions is acknowledged (in (ii)); and it is said that answers to the questions mentioned follow from a consideration of the rules of use of the linguistic expressions concerned. But it is also said (in (i)) that, given the rules of use, 'further explanations of the nature of propositions . . . are theoretically unnecessary', just because 'they follow from the rules: Further,. it is said (in (iii)) 'that such further (extra-systematic) explanations, though 'strictly unnecessary', may be. 'practi.cally us~ful'. in making it easier to learn the use of the expreSSIOns. Fmally (m (IV)) such extra-systematic explanations are said to be analogous to the quasi~ pictorial models which a physicist might use in introducing his theoretical concepts to a beginner. Several points in this deserve comment. (I) The fact that answers to (some) philosophers' questions in some sense follow from the rules of use of the expressions concerned does not have the consequence that it is 'strictly speaking, unnecessary' to give these further explanations, unless one assumes that it is, strictly speaking, unnecessary to take philosophers' questions seriously. For one thing may well follow from another and yet someone may fail to see that it does, unless' it is pointed out to him by means of 'further explanati~ns'; and if this is the situation in this case, then the further explanations are, strictly' speaking, necessary if the aim is to be achieved of resoh,ing the puzzles, of showing how the answers to the conceptual questIOns are implicit in the rules of use of the expressions concerned. (2) Carnap admits that extra-systematic explanations. may nevertheless be useful: he says they may be practically useful in helping someone to learn the use of the expressions concerned. But of course it. i~ characteristic of philosophers' perplexities and questions that they are felt and raised by people who know very well how to use the expressions

509

510

.P. F. STRAWSON

CARNAP ON CONSTRUCTED SYSTEMS

Formal Constructions and Philosophical Understanding

I use the case only ·to bring out (what has not hitherto appeared) that;

Now I want to consider once again, but this time with a different purpose, the earlier part of the passage I quoted. I have tried to show how the passage can be used to explain in part how a certain extreme conception of clarification might come to be held. I now wish to show how it also points to a-less extreme conception, and thereby to a still open issue between constructionism and the analysis of ordinary language. I noted that Carnap acknowledges the existence of typical philosophical questions about, in this case, propositions; and claims that they could be answered5 by attention to 'the system of rules for the linguistic oxpressions of the propositional framework', the system of rules, that is, whereby the framework was 'introduced'. Now it may strike us that in advance of the explicit framing of a system of linguistic rules, there already exists in unfonnalised discourse an ordinary linguistic practice which might itself be said to constitute a propositional framework. That is to say, we commonly use quite a large range of substantival expressions which can occur as grammatical subjects of such grammatical predicates as 'is true', 'is incompatible with so-and-so', etc. These expressions will include clauses beginning with the word 'that'; and also expressions beginning 'the statement that .. .', 'the suggestion that ... " 'the belief that .. .'; and also descriptive phrases which do not incorporate a 'that'-clause, like 'what you said just now', 'what X believes' and so on. Moreover, a comparison of the typical uses of these expressions with those of expressions used to refer to (designate) mental occurrences or linguistic entities or states of a person will 'show that the expressions in question are in fact used differently from expressions of any of these other classes; or, in other words, that the entities which the expressions in question are used to refer to cannot be identified with entities of any of these other classes. So we have, in ordinary unformalised discourse, something very like Carnap's framework of propositions. (We could not conveniently get on without it. And it is merely to echo the main thesis of Carnap's article to add that this is no reason either for despondency or for elation). Here, then, we have a (perhaps untypically) simple instance of an apparent choice of methods. Carnap claims that we can very easily read off answers to (some) typically philosophical questions from a study of the rules of the constructed system..On the other hand, it seems that the same or similar questions can be answered by the examination of the linguistic practice which precedes construction. Why should either method be preferred to the other? I do not propose to debate the general issue on this narrow ground. 5Notice that to deal with the philosophical worry which makes the questions seem 5'0 urgent, more is required than the answers. Carnap provides' something on the necessary lines in the article as a whole.

511

after all, the two methods can, up to a point, be represented as different ways of attaining the same or similar ends. Only if this is so can there b~ an, issue, between them as methods. But it- is important to see how dIffe~ently ,;e must now conceive of the formalist programme of clarificatIOn. It IS no jonger a matter of replacing an unclear concept used for one (non-scientific). purpose with a clear, though related, concept used for a dIffe~ent (sCIentIfic) purpose. The constructed propositional £rame~ork may mde~d .be used by Carnap in attacking other problems; but thIS further use IS trrelevant to its success or failure in the task of clarifying this p~oblem. Unformalised concepts are to be clarified by forma~ constructIOn; and the fact (if it is a fact) that the formal constructIon may then be put to work in new ways, is not now to be taken ; as germane to the purpose of clarification (of this piece of clarification) at all, but as, an extra gift of fortune. Even if agreement can be reached on a common aim of understand~ ing ordinary concepts, however, the danger of trivialising the issue is not altoge.the: a~erted. Let me state a little more fUlly the position the constructIonIst IS now assumed to occupy. He is now to be seen not as offering his construction on the ground of its value for other purposes, nor ~s one who seriously maintains that his system of well-regulated expreSSIOns c~uld actually disp~ace ordinary usage for ordinary purposes. ~e offers hIS system as an object of contemplation which has the follow' lng feature~: first,. it is intrinis~cally clear, in that its key concepts are related m precIseand dete~mmate ways (which the system exhibits), whereas, ex hypothes., the ordmary concepts to be clarified do not have such precise and determinate relations to each other or to other ordinary concepts in terms of which one might seek to explain them; and, second, at least some of the key concepts of the system are, in important respects, very close to the ordinary concepts which are to be clarified. (The qualification 'at least some' is introduced to a110w for the fact that the constructed system may legitimately accord a central place to new concepts whIch do not have any ordinary correlates, but which possess considerable power of unifying or systematising those elements of the system which do have ordinary correlates.) The system as a whole then appears ~s a pre.cise and rigid structure to which our ordinary conceptual eqUIpment IS a loose and untidy approximation. The way in which the debate could once more reach an uninteresting deadlock is the fOllowing. It could be maintained dogmatically on the one hand that nothing but the mastery of such a system would really be understanding, in a philosophical sense, the concepts to be clarified; and to one who maintained this, phrases like 'the underlying logical structure of our con· cepts' might seem to carry the weight of his conviction. Or it might be

P. F. STRAWSON

CARNAP ON CONSTRUCTED SYSTEMS

maintained dogmatically on the other hand that since, ex hypothesi, the ordinary concepts to be examined do not behave in the well-regulated way in which the model concepts of the system are made to behave, there can be no real understanding of the former except such as may be gained by a detailed consideration of the way they do behave, Le. by an investigation of the ordinary uses of the linguistic expressions concerned. Here the deadlock is reached by each party refusing to count as understanding, a condition which is not reached by the method he advocates. There may be something final about this deadlock. For there may here be something which is in part a matter simply of preference, of choice. Nevertheless, there are considerations which may influence choice. For surely, in deciding what to count as philosophical understanding, it is reasonable to remind ourselves what philosophical problems and unclarities are like. Such a reminder I shall briefly attempt in the next section. But I shall partly anticipate it now, in mentioning some general difficulties which arise for -the constructionist in the POM sition he is now assumed to occupy. The constructionist would of course agree that- it is necessary to supply an interpretation for the linguistic expressions of his theory. This is not secured merely by the formal relationships between the constructed concepts which the theory exhibits. At some point it is neces~ sary also to explain the meaning of the linguistic expressions for the constructed concepts in terms which do not belong to the theory and the meaning of which is taken as already known. So some extra~system­ atic remarks are essential. This point need not in itself :raise any particular difficulty. So long as a small number of extra-systematic points ,of contact are clearIymade, the meaning of the rema'ining elements follows from their clearly defined relationships within the system to those to which life has been given by the extra-systematic remarks. (To give a simple instance: it is enough to explain, say, I.' and I,...... in extra-systematic terms-and this is not a hard task-for the interpretation of the remaining constants of the propositional calculus to be fixed.) But if the constructionist claim to achieve clarification is to be vindicated, it is not sufficient:. though it is necessary, that the interpretation of the lin~ guistic expressions of his theory should be determined. For the claim to clarify will seem empty, unless the results achieved have some bear· ing on the typical philosophical problems and difficulties which arise concerning the concepts to be clarified. Now these problems and difficulties (it will be admitted) have their roots in ordinary, unconstrueted concepts, in the elusive, deceptive modes of functioning of unformaIised linguistic expressions. It is precisely the purpose of the reconstruction (we are now supposing) to solve or dispel problems and difficulties so

rooted. But how can this purpose be achieved unless extra-systematic points of contact are made, not just at the one or two points necessary to fix the interpretation of the constructed concepts, but at every point

512

513

where the relevant problems and difficulties concerning theunconstructed concepts arise? That is to say, if the clear mode of functioning of the constructed concepts is to cast light on problems and difficulties rooted in the unclear mode of functioning of the unconstructed concepts, then precisely the ways, in which the constructed concepts are connected with and depart from the unconstructed concepts must be plainly shown. And how can this result be achieved without accurately describing the modes of functioning of the unconstructed concepts? But this task is precisely the task of describing the logical behaviour of the lin- i guistic expressions of natural languages; and may by itself achieve the • sought-for resolutioll of the problems and difficulties rooted in the e!u' sive, deceptive mode of functioning of unconstructed concepts. I should not want to deny that in the diSCharge of this task, the construction of a model object of linguistic comparison may sometimes be of great help. But I do want to deny that the construction and contemplation of such a model object can take the place of the discharge of this task; and I want also to suggest that one thinks that it can, only if one is led away from the purpose of achieving philosophical understanding by the fascination of other purposes, such as that of getting on with science. The point I am making is twofold. First, in so far as the purpose of a constructed system is philosophical clarification, the extra-systematic remarks, so far from being-apart from the minimum necessary to fix~ ing the interpretation-comparatively unimportant trimmings, are just what give life and meaning to the whole enterprise. Second, these extra~ systematic remarks must include exercises in just that method to which system-construction appeared as a rival. ,- ' lVloreover, the general~sefulness of systems of constructed concepts as. objects of comparison with the unconstructed concepts in which our problems are rooted is necessarily limited. For the types or modes of logical behaviour which ordinary concepts exhibit are extremely diverse. To detect and distinguish them is a task in which one may well be hindered rather than helped by fixing one's eye too firmly on the limited range of types of logical behaviour which the concepts occuring in a formal system can there be shown to display. This is not to say that the metavocabulary of description and classification should not itself be made as systematic as possible. (This aim, it need hardly be said, is entirely independent of formal' systematisation of the' concepts: which the metavocabldary is used to discuss.) But (1) an adequate set of meta:concepts for the, dissection of :the expressions of a natural language will scarcely be found by attending primarily to artificial languages; and (2)

P. F; STRAWSON

CARNAP ON CONSTRUCTED SYSTEMS

clarity about the metaconceptsthemselves will be achieved only by attention to the use that is made .of them and hence, ultimately, by at· tention to the' actual functioning of the concepts they are used to discuss. It is the same with-the improvement and refinement of such,metaconcepts. Classifications are found to be crude and misleading, to obliterate logical features, to blur distinctions; and these discoveries, too, are made by attention to the actual modes of functioning of 'actually used linguistic

philosophical problems, difficulties and questions. But at any rate this much will be broadly agreed: that they are problems, difficulties and questions about the concepts we use in various fields, and not problems, di:ffi~ culties and questions which arise within the fields. of their use. (A phil' osophical problem about mathematics is not a mathematical problem.) To say more is to risk the loss of general agreement. Nevertheless, I think iti~ possible roughly to distinguish, though not to separate,. certain strands or elements in the treatment of this diverse mass ofconceptual ques,:" tions: First, and very centrally, we find the necessity of dealing with paradox and perplexity. For it often happens that someone reflecting on a .certain set of concepts finds himself driven to adopt views which seem , to ·others paradoxical or unacceptably strange, or to have c9nseql;lence~ .~ which are paradoxical or unacceptably strange. (He mayor may not f himself embrace these conclusions with complacency.) Or-the obverse of this-it may happen that someone so reflecting becomes unable to see how something that he knows very well to be the case can possibly be the case. In this situation the critical philosopher must not only restore the conceptual balance which has somehow been upset; he must also diagnose the particular sources of the loss of balance, show just how it has been upset. And these achievements are not independent of each other. It also seems to me possible to say in general what kind of thing the source of conceptual unbalance is. Such unbalance results from a kind of temporary one-sidedness of vision, a kind of selective blindness which cuts out most of the field, but leaves one part of it standing out with a peculiar brilliance. This condition may take many different, though interconnected, forms. The producer of philosophical paradox, or the sufferer from philosophical perplexity, is temporarily dominated by One logical mode of operation of expressions, or by one way of using language, or by one logical type or category of objects, or by one sort of explanation, or by one set of cases of the application of a given concept; and attempts to see, to explain, something which is different, in terms of, or on analogy with, his favoured model. The distortions which result froni such attempts are of equally many kinds. To correct the distortions, one must make plain the actual modes of operation of the distorted concepts or types of discourse; and, in doing this, one must make plain the differences between their modes of operation and those of the m.odel concepts or types of discourse; and, in doing this, one must, if . one can~ make plain the sources of the blinding obsession with the model cases. . This, then, is one strand in the treatment of philosophical problems; and I call it central, partly because the need for it has in fact provided so strong an impetus to the whole activity. From it can be distinguished, though not separated; two other strands. One is the attempt to explain, not just how oui concepts and types of discourse operate, but why it is

514

expressions.

Finally, I may suggest that the very success of logicians in developing techniques of fonnalisation has itself generated philosopltical problems which cannot, in their very nature, be solved by further essays in the use of these techniques. This is not, of course,' a reproach to the logicians; It is char.acteristic of major scientific advances that the effective use of the new concepts and methods introduced in making them may precede the adequate philosophical understanding of that use, and hence of the relation of these concepts and methods to others belonging to different, though perhaps overlapping, fields. Descartes' mathematical ideal of knowledge has such a source, and so have the recurrent perplexities about perception which the work of physicists and psysiologists engenders. Nor is it in any way to be regretted that these problems should arise; for their resolution results in a clearer, more self-conscious

understanding of what we are doing both with new concepts and with old. But it is necessarily not within the field of the puzzle-generating advance that such problems as these can be solved. For these problems are defined as those which result from the attempt to make inappropriate applications of the concepts of the field. So may we see in the barely sketched but grandiose plan of logical atomism the outlines of an attempt to find in ordinary empirical discourse the real formal structure which the planners were encouraged by the advances of logic to believe must be there to be found. And so, to set a small thing beside a large one, we may see in the attempts to analyse the ordinary conditional in terms exclusively of the constants of modern elementary logic, the force of the conviction that concepts successful for some purposes must be adequate for others. From such attempts we may learn much; but not by their succeeding. Part of what we have to explain and free ourselves from, in dealing with them, is the undue fascination exercised by fonnal systems.

Philosophy and Ordinary Language 6 It is, no doubt, rash to attempt to describe in general the nature of 6See Professor Ryle's article, "Ordinary Langllage," The Philosophical Review (Apr.il, for a discussion of. this topic, and, in particular. for the removal of same mlS· \1ndenitandings about the. phrase "ordinary language." ~953),

51q

516

P. F. STRAWSON

that we have such concepts and types of discourse as we do; and what alternatives there might be. This is not an historical enquiry. It attempts to show the natural foundations of our logical, conceptual apparatus, in the way things happen in the world, and in our own natures. A form which propositions exemplifying this strand in philosophy may often take, is the following: if things (or we) were different in such-andsuch ways, then we might lack such-and-such concepts or types of discourse; or have such-andMsuch others; or might accord a subordinate place to some which are now central, and a central place to others: or the concepts we have might be different in such-and-such ways. It mIght reasonably be maintained, or ruled, that full understanding of ~ ?~n­ cept is not achieved until this kind of enquiry is added to the actIVItieS of. comparing, contrasting and distinguishing which I mentioned first. Of course speculations of this kind are restricted in certain ways: they are limited by the kinds of experience and the conceptual apparatus we in fact have. But this is only the restriction to intelligibility; it leaves a wide field open to philosophical imagination. The distinction I used above between the way things happen in the world, and our own natures, is here, though vague, important. For it is a part of our nature that, things other than ourselves being as they aTe, it is natural for us t~ have the conceptual apparatus that we do have. But human nature LS diverse enough to allow of another, though related, use of philosophical imagination. This consists in imagining ways in which, without th~ngs other than ourselves being different from what they are, we mIght view them through the medium of a different conceptual apparatus; Some metaphysics is best, or most charitably, seen as consisting in part in exercises of this sort.. Of course, even when it can be so interpreted, it is not presented as a conceptual or structural revision by means of which we might see things differently; it is p:esented as a pietu.re of thing~ as they really are, instead of as they delUSively seem. ~d thIS .pres~ntatIOn, with its contrast between esoteric -reality and dally delUSIOn, Involves, and is the consequence of, tp.e unconscious distortion of ordinary c?ncepts, i.e. of the ordinary use of linguistic expressions. So metaphYSICS, though it can sometimes be charitably interpreted in the way I suggest~ in fact always involves paradox and perplexities of the kind I first mentioned; and sometimes embodies no rudimentary vision, but merely rudimentary mistakes. 7 There is a third strand to be distinguished; something soberer than 7It might seem that in the foregoing I have committed myself to an unintellig~ble notion of things as they are as opposed to things ,as we see them through the medlUm of a conceptual apparatus. But to think this is to forget ,that I have made ~se of a rough distinction between things, ?the:r,- t~an ou~elves. and o~r,?wn ~atures, .Interests and' needs. Features of each' can mtelliglbly be unagmed to vary-with varymg conteptuaJ consequences-while the other remainsc6nstant.

CARNAP ON CONSTRUCTED SYSTEMS

517

the second. That examination of current concepts and types of discourse to which paradox and perplexity so commonly give the initial impulse, can be pursued with no partiCUlar therapeutic purpose, but for its own sake. This is not to say that puzzlement is not in question here. One can, without feeling any partiCUlar temptation to mistaken assimilations, simply be aware that one does not clearly understand how some type of expression functions, in comparison with others. Or, having noticed, or had one's attention drawn to, a certain logico~linguistic feature appearing in one particular area of discourse, one may simply wish to discover how extensive is the range of this feature, and what other comparable features are to be found. Of course, the resulting enquiries ~ may well pay therapeutic dividends. But this need not be the purpose for which they are undertaken. . . In relation to the first and third of these three types of philosophical aIm: the .roles of the two apparently contrasted methods of philosophical clanficatIOn should already be clear. The description of the modes of functioning of actually employed linguistlc expressions is of the essence of the third aim; and it is simply the least clouded form of a procedure which is essential to the achievement of the first. Here the arguments of the previous section apply. To observe our concepts in action is necessarily the only way of finding out what they can and cannot do. The right kind of attention to the ordinary use of expressions provides a means of refutation of theories founded on mistaken assimilations; it .provides a description of the actual functioning of the problematic concepts, to take the place of the mistaken theory; and, finally, it helps, or may help, with the diagnosis of the temptations to the mistakes. This last it may do because the analogies which seduce the philosopher are not, in general, private fantasies; they have their roots in our ordinary think.:. ing: and show themselves in practically harmless, but detectable ways, in ordmary language-both in its syntactical structure and in the buried figures .which individual words and phrases contain. I have already acknowledged that system-construction may have an ancillary role in achiev~ ing these two types of aim, and given reasons for thinking that it mu'st remain ancillary-and limited. Model objects of linguistic comparison may h~lp us to understand the given objects; but it is dogmatism to mamtam that the construction of model objects is the best or the only means of achieving such understanding. S With the second philosophical strand I distinguished, the case is somewhat different. To understand the foundation of our concepts in natural facts, and to envisage alternative possibilities, it is not enough to have a sharp eye for linguistic actualities. Nor is system-construction a direct contribution to the achievement of the first of these two, i.e. to

J

BCf. T.ke Logical Syntax ot Language, Introduction, 8.

518

P. F. STRAWSON

seeing why we talk as we do. But it may be the second, i.e. to imagining how else we might talk. The constructionist may perhaps be seen as an enlightened metaphysician-one who, perhaps wistfully, envisages the possibility of our situation and our need for communication so changed and simplified that such a well-regulated system of concepts as he supplies is well adapted to both. It is only when the claim to exclusiveness is made on behalf of the constructionist method, and of particular constructions, that one must begin to query the enlightenment. For behind these claims may lie a formalizing mystique: the belief that the model systems embody the real structure of our concepts, hidden from us by the untidiness of our actual practice. But, again, this claim may be softened to the expression of a preference-which leaves one no more to say. To conclnde, then. There is not just one thing which is legitimately required of the philosopher who would increase our conceptual understanding. In particular, it is certainly not enough to say that he should describe the functioning of actually employed linguistic expressions. For simply to say this would not be to give any indication of the sort of decription he should provide. That indication is given when it is shown ~ow description of the right sort may bear upon our conceptual confusIOns and problems. Next we see how more may be required of. hi:n than the resolution of these confusions with the help of those descnptIOns; how a more systematic classification and ordering of the types of discourse and concept we employ may be sought; how a fuller understanding of both may be gained by enquiring into their foundation in natural facts; and how room may here be found for the envisaging of other possibilities. If the philosopher is to do all or only some of these things, it is true that he cannot stop short at the literal description, and illustration, of the behaviour of actually used linguistic expressions. Nevertheless, the actual use of linguistic expressions remains his sole and essential point of contact with reality; for this is the only point from which t?e ~etual mod~ of operation of concepts can be observed. If he severs thIS VItal conneXlOn, ~Il his ingenuity and imagination will not save him from lapses into the and or the absurd.

P. F. UNIVERSITY COLLEGE OXFORD UNIVERSITY

17 Yehoshua Bar-Hillel

REMARKS ON CARNAP'S LOGICAL SYNTAX OF LANGUAGE

A FTER I had accepted the task of evaluating Carnap's Logical Syn. £l.. tax of Language 1 for the present volume, I cherished for some time the thought of both presenting the main ideas of the Logical Syntax and of criticizing them in the light of the progress made in logic and methodology during the last twenty years. But one more careful reading of the book made me realize the absurdity of my original intention. How could one possibly summarize, and critically evaluate, the contents of a book in a few dozen pages, when every single one of its sections contains such a wealth of ideas, painstakingly elaborated, carefully explained and illuminatingly illustrated? Not all of these ideas were original with the author, but even when he adopted somebody else's flashes of genius-his debts to Frege, Russell, Wittgenstein, Hilbert, Godel and Tarski are acknowledgedly great-he made them change their character and often gain in importance by incorporating them into his own general framework. How would one go about condensing a book when he is convinced that often not a single word can be omitted, not a single illustration discarded, not a single historical aside passed over without becoming involved in some serious 1055 and when he has, moreover, every few pages the impression that the author could and should have said much more on a certain subject and that only lack of space prevented him from giving us the enlightenment for which we now have to struggle all by ourselves. There are many pages containing short remarks that carry convincing proof that Carnap must have deeply thought about the problem treated there but would have needed many more pages to expand 1

1

STRAWSON

IThe following abbreviations will henceforth be employed:· LSL for The Logical Syntax ot Language (London and -New York, 1937), being an expanded and corrected translation of the German original Die logische Syntax der Sprache. (Vienna, 1934); TM for "Testability and Meaning," Philosophy ot Science, III (1936), 419-471, and IV (1937), 1·40, reprinted by Graduate Philosophy Club, Yale University, New Haven, Connecticut (1950); ESO for "Empiricism, Semantics, and Ontology," Revue Intern.ationale de Philosoph ie, IV (1950), 20·40, reprinted in Readings in Philosophy ot Sci· ence, ed. P. P. Wiener (New York, 1953), 509-522 (and quoted according to this reprint).

519

;